CIRED
19th International Conference on Electricity Distribution
Vienna, 21-24 May 2007
Paper 0638
A COMPARISON OF GROUNDING TECHNIQUES FOR DISTRIBUTED GENERATORS
IMPLEMENTED IN FOUR-WIRE DISTRIBUTION GRIDS, UPS SYSTEMS AND
MICROGRIDS
Annick DEXTERS
KHLim – Belgium
annick.dexters@khlim.be
Tom LOIX
K.U. Leuven - Belgium
Johan DRIESEN
K.U. Leuven – Belgium
Ronnie BELMANS
K.U. Leuven - Belgium
tom.loix@esat.kuleuven.be
johan.driesen@esat.kuleuven.be
ronnie.belmans@esat.kuleuven.be
ABSTRACT
This paper starts with an overview of the common
grounding techniques used in distribution grids. Next some
important issues and problems related to grounding of
distributed generators implemented in these grids will be
described. Two applications will be discussed in detail:
UPS systems and microgrids. The presence of a neutral
conductor in these systems and the ability to operate both in
grid-connected and islanded mode complicates the
grounding system design. The grounding requirements need
to allow correct operation in each of the two operating
regimes.
INTRODUCTION
Electric power systems require two types of grounding:
system grounding and equipment grounding. The function
of system grounding is to limit voltages due to lightning,
line surges or unintentional contact with higher voltage
lines, as well as stabilizing the voltage to ground during
normal conditions. Equipment grounding must assure a low
potential difference between nearby metallic frames to
protect people from electric shock and protect property from
being damaged, and facilitates the operation of circuit
protective devices for ground fault currents.
The deregulation of utilities, the emerging power markets
and the Kyoto protocol all tend to increase the penetration
of distributed generation (DG). Small DG units are
connected at the distribution level of the utility system near
the customers. Introducing additional generators at this level
causes a redistribution of both load and fault current, and
adds a possible source of overvoltages. The utility circuits
were designed as a radial system, in which fault clearing is
achieved by opening only one protective device because
there is only one source contributing to the fault. In the
presence of DG, there are multiple sources and opening of
the utility breaker only does not guarantee that the fault is
cleared. Therefore, DG is often required to disconnect from
the system when a fault is detected in order to revert to a
true radial system and be able to proceed with the normal
fault clearing procedure. It is possible that the DG units
disconnect either too soon or too late, resulting in
detrimental impacts on the distribution system. In both
cases, there are potential operating conflicts with respect to
overcurrent protection and voltage restrictions [1].
CIRED2007 Session 2
Paper No 0638
THE COMMON GROUNDING TECHNIQUES
IN DISTRIBUTION GRIDS
In case of a single power source, grounding is relatively
straightforward. In radial distribution grids this condition is
fulfilled. A grounded distribution system is usually derived
from a distribution substation transformer with Y-connected
secondary windings. The neutral point of the windings can
be solidly grounded or connected to ground through a
current-limiting device such as a resistor or reactor.
Alternatively, a grounding transformer may be used to
establish a grounded system. In four-wire distribution
systems, the neutral conductor is either connected to earth
several times (multigrounded), which is characteristic for
many MV grids in US and LV grids in Europe, or fully
insulated, being only connected to earth at the source
(unigrounded). In three-wire unigrounded systems, e.g. the
MV grids in Europe, a neutral conductor is not run with
each circuit, but the system is grounded through the
connections of the substation transformer or grounding
transformer.
Each method of grounding has its own advantages and
disadvantages. One can roughly state that as the single
fault-to-ground current decreases due to current-limiting
devices, the overvoltages on the unfaulted phases reach a
higher value and the sensitivity of the fault detection must
be increased. These overvoltages can result in power quality
problems and failure of arresters [2].
Systems rated 600V or less are always solidly grounded
to facilitate overcurrent device operation in case of
ground faults. The prevailing practice in US is to connect
the utility neutral to the equipment ground of the
customers. In Europe on the other hand, one distinguishes
the TT, TN and IT systems where the utility neutral and
the equipment ground are not necessarily bonded (IEC
60364).
GROUNDING OF DG UNITS
An important share of the present DG units uses a
synchronous generator. One must pay attention when
choosing the type of grounding, since it is not always
recommendable to solidly ground the neutral of the
generator.
In case of a stator ground fault extensive damage can occur
to the generator, even after opening the circuit breaker. This
damage is due to the time required for the excitation field to
Page 1 / 4
CIRED
19th International Conference on Electricity Distribution
Vienna, 21-24 May 2007
Paper 0638
decay, thereby maintaining the flow of current to the fault
Third harmonic currents, driven by the third harmonic
voltage inherent to the generator, can circulate in the power
system zero sequence network. The magnitude of these
currents can be reduced by a proper neutral grounding
device or a generator with a 2/3 winding pitch.
The ground fault current from a solidly grounded generator
is larger than the three-phase fault current since the natural
zero sequence impedance of a synchronous generator is
typically about half the subtransient positive sequence
reactance [3]. Therefore the appropriate technique of
grounding standard generators is inserting an impedance
between the neutral point and earth or grounding the
interconnection transformer.
Several winding arrangements for interconnection
transformers are possible: ∆-grounded Y, ∆-∆, grounded Y∆ and grounded Y-grounded Y, each of which has its
advantages and disadvantages, depending on the grounding
technique of the utility. Here the focus is on the fault
contribution and overvoltages for DG units coupled to a
four-wire multigrounded distribution grid.
Figure 1 shows several single line-to-ground (SLG) fault
locations. In case the primary winding on the utility side of
the interconnection transformer is not grounded, the DG
unit does not supply any ground fault current for a fault at
F1 or F2 (at least when the DG is decoupled before the
utility breakers are opened). If the primary winding is a
grounded Y, the DG unit can backfeed the distribution
feeder, which might be a serious problem, certainly in fourwire multigrounded distribution grids.
A grounded Y-∆ transformer creates additional ground
current paths as is depicted in Figure 2. The fault currents
are no longer flowing just in one path from the substation to
the fault, but in several parallel paths even downstream of
the fault. This can result in malfunctioning of protective
relays and needlessly blowing line and transformer fuses.
One common side effect is that the feeder breaker will trip
for any SLG fault on all feeders served from the same
substation bus. The utility transformer might be
overstressed repeatedly and eventually fail. In order to
prevent the sympathetic tripping of the feeder breaker for
SLG faults on other feeders directional overcurrent relays
can be used [1].
To understand how a DG unit can create an overvoltage
during ground fault, consider an interconnection
transformer with a ∆ winding on the utility side as depicted
in Figure 3. A permanent SLG fault has occurred and the
Figure 1: Overview of SLG fault locations
CIRED2007 Session 2
Paper No 0638
Figure 2: SLG fault contribution of a DG unit
utility interrupting device has opened before the DG was
disconnected. This leaves an isolated system energized by
the DG unit. There is no longer a grounded source on the
utility side of the transformer. Now the potential of the
neutral essentially equals that of the faulted phase. Any
loads or voltage arresters connected between an unfaulted
phase and neutral are subjected to a line-to-line voltage, as
is shown in Figure 3. This could damage the loads after just
Figure 3: Load voltage rating violation
a few cycles. However the overvoltage magnitude decreases
significantly in case the DG unit is not strong enough to
feed the islanded network [4].
Using interconnection transformers is the prevailing
practice for connecting medium-power DG units in parallel
with the MV distribution grid. Low-voltage synchronous
generators on the contrary may be solidly grounded not only
in island mode but also in grid-connected mode They
typically appear to have sufficient bracing and a 2/3
winding pitch to permit this. Solidly grounding is however
not possible for asynchronous generators because of third
harmonic currents associated with this type of generator.
However, the main considerations formerly discussed are
also valid in case of a direct connection of these generators
to the utility grid.
GROUNDING OF UPS SYSTEMS
Installing a UPS system is the most common way to ensure
a very reliable and high quality power supply to sensitive
loads, such as computer data centers or telecommunication
centers. These kinds of loads are very sensitive to
disturbances of the power supply parameters, mainly the
voltage amplitude and frequency, from their rated values as
well as a (temporary) loss of power supply. Using a UPS
system for these loads enables to protect these loads against
grid disturbances and ensures continuous operation of the
loads, avoiding expensive shutdowns and restarts.
Page 2 / 4
CIRED
19th International Conference on Electricity Distribution
Vienna, 21-24 May 2007
Paper 0638
One important aspect of designing a UPS system is
grounding. When the grounding is poorly implemented,
serious problems might arise, resulting in seriously
degraded performance, which is quite the opposite of what
one has in mind when installing a UPS system. The
discussion of grounding techniques for UPS systems given
below focuses on on-line static UPS systems. However, the
main considerations and conclusions are also applicable to
rotary UPS systems.
It is important to note that an on-line UPS system usually
has at least two different operation modes: one in which the
loads are fed by the UPS power electronic converter(s), the
other in which the loads are fed by the utility grid (when the
UPS bypass is closed). Implementing a grounding strategy
operating well in both modes is necessary to provide a high
level of power supply to the loads at all times.
On-line static UPS system topology
Figure 4 shows the UPS system topology used in this paper
to outline the main grounding issues and solutions for UPS
systems. The building including sensitive loads is connected
to the medium-voltage utility grid using several
transformers. Also emergency diesel generators are
installed, able to feed the loads in case of a black-out in the
utility grid. The transformers and emergency generators are
connected to two distribution busbars, as well as the input
rectifiers of the UPS systems and the static (and possibly
also manual) bypass switches. The output of the UPS
inverters is connected to an LC-filter and subsequently to a
Static Transfer Switch (STS). It is also possible to connect
the UPS system outputs to several load distribution busbars,
which in turn are connect to the STSs. Each group of loads
is connected to the output of an STS, being able to be fed by
two parallel and independent UPS systems (and possibly the
utility grid, in case the “active” UPS system works in
bypass mode).
When choosing the type of grounding system for a UPS
system, the IT system is sometimes preferred because of its
enhanced continuity of power supply, even during the
presence of the first fault. The fault has to be found and
eliminated before a second fault occurs. Here a TN-S
grounding system is chosen, which is often preferred for
installations with an important amount of (power) electronic
loads and associated EMI filters [5].
Main grounding issues for a UPS in a TN-S system
The main issue regarding the grounding of UPS systems is
the connection of the neutral conductor to earth. It is crucial
to make this connection only once, otherwise an undesirable
intermingling of ground and neutral currents might occur,
resulting in unwanted voltages, which are capable of driving
common mode currents, and incorrect ground-fault sensing.
This is completely the opposite of the prevailing practice of
multigrounding the neutral of four-wire distribution grids.
CIRED2007 Session 2
Paper No 0638
Figure 4: On-line static UPS system
Figure 4 shows two points at which the neutral conductor is
connected to earth: one at the neutral point at the secondary
side of the transformers (as well as the emergency
generator), upstream of the UPS system converters and one
at the output of the converters. While the first point
provides an earth link in case the loads are fed by the utility
grid (through the static bypass switch), the second one
provides an earth link when the loads are fed through the
power electronic converters. It is vital to use 4-pole static
bypass switches and 4-pole switches at the output of the
converter filter, in order to avoid a double connection of the
neutral conductor to earth, which would result in a loop
formed by the neutral and earthing conductors. It is also
important to use 4-pole STSs in this topology [5],[6]. In
case one just connects the neutral conductors of the two
parallel sources feeding the STS (as is done in a 3-pole
STS), another neutral and ground conductor loop appears.
Simulation results
This section describes some simulation results, obtained
using Matlab®-Simulink and the PLECS toolbox,
concerning a UPS topology similar to the one shown in
Figure 4. The main differences between the topology used
for the simulations and the one depicted in Figure 4 are the
use of three-pole static bypass switches, effectively
connecting the upstream neutral conductor with the neutral
conductor started at the secondary side of the inverter
output transformers, the use of a three-pole STS. This
Page 3 / 4
CIRED
19th International Conference on Electricity Distribution
Vienna, 21-24 May 2007
Paper 0638
situation is actually witnessed in several buildings. The load
varies between a balanced total of 96 kVA and an
unbalanced total of 160 kVA.
An important issue illustrated here is the occurrence of stray
currents, due to the existence of several parallel paths for
the neutral current, which might be large due to the
important share of single-phase loads and the amount of
unbalance associated with them. Figure 5 shows the
currents through two neutral conductors in case the load is
varied one phase at a time. Inverter A (see Figure 4),
supplying the loads and its neutral current is shown in the
uppermost part of Figure 5. The current through the neutral
conductor of inverter B (Figure 4), connected to the same 3pole STS, is shown in the lower part of Figure 5. One can
see that an important amount of the neutral load current
returns through the neutral conductor of inverter B and the
earth conductor connecting the neutral point of inverters A
and B. There are even more possible paths for the neutral
load current, for instance through the neutral conductors
starting at the distribution busbars, which are always
connected to the neutral conductors at the inverter output in
case one uses 3-pole static bypass switches and no isolating
transformer in the bypass channel, and through the
grounding conductors.
GROUNDING IN MICROGRIDS
The MicroGrid concept [7] has received much attention as
it offers a way to both support the growing penetration of
distribution generation and increase the reliability and
continuity of electricity supply to customers. A microgrid
consists of a group of local (distributed) generators and
loads (e.g. a city district or an industrial estate), which are
normally connected to the utility grid, but are also able to
operate in island mode (disconnected from the utility grid).
The MicroGrid must achieve the same level of safety as any
other conventional distribution system in both modes.
According to current legislation, LV generators may be
earthed or unearthed when operating in parallel with the
distribution system. The usual practice is not to earth the
generator neutral point in parallel operation because the
earthing of several generators dispersed in the MicroGrid
hampers control of earth fault currents, detection of earth
leakage current and avoiding interference to communication
systems. In that case an earth reference point must be
provided when the MicroGrid is disconnected from the
main grid. One possible solution is to operate a generator
with an unearthed star point when grid-connected and then
automatically reconnect the star point to earth when the
MicroGrid is islanded. A more evident solution is using the
source earth of the distribution transformer. So, a MicroGrid
should be disconnected from the main grid only by opening
the circuit breaker upstream from the transformer. Then the
micro-sources could be operated safely without earthing
their neutral points locally [8].
REFERENCES
[1]
[2]
[3]
Figure 5: Neutral currents in case of unbalanced load
The presence of stray currents might cause malfunctioning
ground-fault protection (GFP), in case the protective
devices monitor the three phase conductors and the neutral
conductor. The part of the neutral current returning through
the grounding system is not seen by the GFP, so the system
might be shut down because the GFP mistakenly suspects
the presence of an earth fault. Conversely, in case of an
earth fault occurring on one of the live conductors, part of
the fault current might return through the neutral conductor.
The GFP “sees” a smaller fault current than the one actually
present and might decide not to open the circuit breakers.
It is clear that both safety and continuity of supply might be
significantly reduced when using 3-pole STSs and static
bypass switches in a UPS topology with a TN-S grounding
system. Application of the structure depicted in Figure 4
allows eliminating these errors and guarantees correct
operation under all circumstances.
CIRED2007 Session 2
Paper No 0638
[4]
[5]
[6]
[7]
[8]
R.C. Dugan, T.E. McDermott, “Distributed generation”,
IEEE Ind. Appl. Mag., March/April 2002, pp. 19-25.
J. Burke, M. Marshall, 2001, ”Distribution system neutral
grounding”, Proc. of the IEEE Transm. and Distr. Conf.
and Exp., vol.1, 2001, pp. 166-170.
P. Pillay et al., ”Grounding and ground fault protection of
multiple generator installations on medium voltage
industrial and commercial power systems- Part 2:
Grounding Methods”, IEEE Tr. on Ind. Appl., vol. 40,
No. 1, January/February 2004, pp. 17-23.
P. Barker, “Overvoltage Considerations in Applying
Distributed Resources on Power Systems”, Proc. Of the
2002 IEEE PES Summer Meeting, vol. 1, pp. 109-114.
J.-N. Fiorina, “Uninterruptible Static Power Supplies and
the Protection of Persons”, Cahier Technique no. 129
(Collection Technique de Schneider).
H. O. Nash, “More about Standby Generator Grounding,
GFP, and Currents that Go Bump in the Night”, IEEE Tr.
on Ind. Appl., vol. 33, No. 3, May/June 1997, pp. 593600.
R. H. Lasseter and P. Piagi, “Microgrid: A Conceptual
Solution”, Proc. of the 35th Annual IEEE Power
Electronics Specialists Conf., Aachen, Germany, June 2025, 2004, pp. 4285-4290.
N. Jenkins et al., “Safety Guidelines for a MicroGrid”,
Deliverable DE1, Internal report for MicroGrids project,
Nov. 2004.
Page 4 / 4